These computer voices sound human enough to mislead, but one layer of speech still breaks the illusion

A Max Planck Institute study in Frankfurt found that human perception of voices depends on three factors: acoustic characteristics like timbre and intonation, sentence content, and listener's language comprehension.